When deploying custom knowledge models for business operations, data isolation is not a feature—it is a compliance requirement. Business executives must know their files are fully protected and never leaked.
At VeloReply, we utilize Postgres with secure pgvector extensions to handle high-dimensional vector embeddings. This engineering breakdown details our security audit controls, isolated token access keys, and server pipelines.
1. Multi-Tenant Database Isolation (MTDI)
We employ strict Row-Level Security (RLS) policies at the database layer. Every table query enforces an explicit, indexed tenant key block. An administrative tenant can never read, query, or map vectors belonging to another company configuration.
Multi-Tenant Security Standards
- AESA-256 Vector Keys: Document shards and generated embeddings are encrypted in transit and at rest.
- Ephemeral Memory Processing: Uploaded files are segmented, vectorized, and purged from local system directories instantly.
2. Server Notifications & Safety Whitelisting
To safeguard against phishing, all critical system activities—including login alerts, vector generation complete triggers, and password resets—arrive strictly from:
Ensure your organization's corporate firewalls whitelist this verified sender address to prevent missing important model sync alerts.